Originally posted by: Philippart
I love to undervolt, I don't rely on stress testing software, I run distributed computing projects in the background and the pc is always on. If it doesn' crash during normal use+dc in the background for half a week, I consider it as stable, if it happens to crash one day I can still raise the vcore 1 step.
What do I care about a stress test if my pcs run stable for my use 24/7 for years?
EDIT: and yes dc is in a sense a stress test too, but less aggressive and more like normal programs you would use every day, not specially designed to "torture" your hardware.
Hence why if I use DC to set "stability" of my system I would expect it to run 1-2 weeks without a burp, EEU, whatever....before I flung out the words "its stable"...
On the other hand that is a lot of investment to have it fail 4,5, or 6 days into it. So I run prime95 instead for 24-48 hours, and Memtest for 8-12 hours. It is above the level DC will ever reach with it in normal use , but that is what is great about weeding it out. I have never had a system that was stress tested like I mention above fail in a DC program.
This is the reason why when I set an OC I dont have to 3 months down the line tick it down a notch or raise the vcore a notch. I see this a lot. Those systems or components were iffy in the beginning if that happens, and usually because they lacked stress testing it thoroughly enough.
I also dont overclock my memory....just a rule of mine and it keeps the problems down. In the past I have seen memory good bad much faster then any cpu I have ever oc'd. Memory stable now can be unstable in a few months. I have never seen that with a cpu I have owned.